The article presents a new method called crowdsourced adaptive survey (CSAS), which combines natural language processing and adaptive algorithms. The method transforms open-ended text from participants into potential survey questions, which are then prioritized using a multi-armed bandit algorithm. This allows the survey to evolve over time, capturing emerging issues and claims. The method proves effective in identifying salient issues within minority communities and evaluating issue importance in the aggregate. It showcases potential in studying topics where participant-generated content may enhance our understanding of public opinion.
Publication date: 25 Jan 2024
Project Page: https://arxiv.org/abs/2401.12986v1
Paper: https://arxiv.org/pdf/2401.12986
